一种用于IC内部焊点识别的快速混合算法

A Fast Hybrid Algorithm for Recognition of IC Inward Solder-Joints

针对IC内部的焊点匹配定位问题,提出了一种从粗到精的快速混合算法。该算法首先针对目标焊点分布在IC芯片边缘附近的特点,通过基于阀值自动选取的图像分割算法来确定出焊点目标的潜在区域,再基于图像几何矩的快速计算方法,在潜在区域内采用中心矩匹配和序贯相似性检测(SSDA)相结合的策略来精确定位微小焊点。实验结果表明,该混合算法在不失匹配精确度的条件下,使匹配速度得到显著的提高,从而满足了实时图像识别系统的性能要求。

A fast hybrid algorithm of from coarse to fine is presented for the problem of matching and location of IC inward solderjoints. First considering the feature that target solder-joints located in the vicinity of the edges of an IC chip, its potential area can be detected by an image segmentation algorithm based on threshold automatic selection. Then, based on a fast calculating method of an image＇s geometric moments, a combined method of centralized moments matching and SSDA is applied to determine the target＇s precise location in potential area. Experiment results show that the hybrid algorithm can observably improve matching rate without the loss of matching accuracy and satisfy the performance demand of real-time image recognition system.